Quality Analysis of Tomato for Ketchup Manufacturing Industry Using Deep Learning

Maintaining the product quality is one of the most essential techniques when it comes to food manufacturing industries. Most of the food industries generally followed traditional and manual methods for sorting the raw materials. But nowadays industries are using various high-end technologies. One of the most beneficial technologies are the machine learning and deep learning techniques as they have created a huge impact towards the food industries throughout the manufacturing process with its various applications. Here, the proposed work is based on tomato quality analysis for ketchup manufacturing using a deep Learning algorithm. Where, tomatoes are analyzed based on its external characteristics such as shape, size, color and projections. The process is done using a real time object detection method named YOLO, which uses neural network for the real time object detection. The dataset taken for training the model are real time tomato images based on different conditions. Once the custom dataset is created the model obtained is trained and tested using various deep learning algorithm named YOLOv3, YOLOv5 which uses convolution neural network to identify the objects. The outcome of the trial indicates that YOLOv5x has the highest precision and highest accuracy amongst all the other algorithm, where the obtained Mean average precision at 0.5 of the model is 98.7 and the value of the precision is 97.0 and the recall value of the model is 97.9. The obtained results signify that the created model executes accurately and was precisely able to predict the good tomatoes and rotten tomatoes. By using the above techniques in the food industries both quality of the ketchup can also be maintained, and the production of the ketchup can also be increased.